Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add juxt/claude-plugins --skill drivegit clone --depth 1 https://github.com/juxt/claude-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/juxt/claude-plugins/drive)<a href="https://agentmods.dev/skills/juxt/claude-plugins/drive"><img src="https://agentmods.dev/badge/skills/juxt/claude-plugins/drive/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/juxt/claude-plugins/drive"><img src="https://agentmods.dev/badge/skills/juxt/claude-plugins/drive.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00166 | $0.01691 |
| Opus 5.5 | $0.00066 | $0.00676 |
| Sonnet 5.5 | $0.00033 | $0.00338 |
| Haiku 4.5 | $0.00017 | $0.00169 |
Grade A, and why
drive scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 106 lines — stays where its author put it; the contents beside it link to each section on GitHub.
a-star Drive
Interpret MUST, MUST NOT, SHOULD, SHOULD NOT, MAY, etc. per RFC 2119.
Success: a small change, ready to merge, that the user can review from its diff and message alone — and, once it lands, page that part of the card out of their head.
A dart takes the straight line to the goal; a drive takes the road. The walls the dart ignored — the project's coding standards, its tests, its definition of done, its code review — are what a drive navigates.
Load a-star:a-star first.
Modes
Every drive is a tidy or an advance, decided before it starts, and the two never share a commit.
-
Tidy — changes structure, never behaviour. Success: an equivalence change that gives us options, reviewable by asking only could this change anything? A change that could change anything isn't a tidy.
-
Advance — changes behaviour, moving the base a step towards the goal. A structure change turning up mid-advance is a tidy, never part of this commit: decide now, later or never, and if now, it interrupts the advance.
In either mode
- One ready sub-goal, small enough for a reviewer to hold in their head; one too big is two — record both before starting.
- What the darts found is evidence, never code to edit into shape.
- Write it to the project's coding standards.
- Obviously no bugs, not no obvious bugs — Hoare's distinction. The test: a human reviewing the change in isolation can see it has none — for a tidy, this is its success criterion. Where they can, they confirm it and page it out, which takes far more off their mind than any review that leaves them unsure they've looked hard enough. A change that fails the test wants a tidy under it first, or is two sub-goals.
- Simple, not easy — Rich Hickey's Simple Made Easy. Un-braid concerns rather than reach for what's nearest to hand; state that can be derived, or can drift from what it mirrors, shouldn't exist. The test: a human reviewing the change in isolation can see each concern on its own, confirming one without holding the rest of the change in their head.
- No generality the sub-goal didn't ask for.
- No dead code. Everything the commit adds has a caller outside the tests in this commit; a new abstraction nothing yet calls lands with the first change that calls it. A tidy reshapes code that already runs.
- The suite stays green. A test weakened or disabled to get past it is a broken invariant: report it, don't commit it.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago Changed · +13 lines 4232bc556aec
- 4d ago Changed · +5 lines 0a16e65e12a2
- 7d ago Changed · +46 lines · +87 tokens per session 31c4bf688661
- 9d ago Changed · +1 lines 811dee0cfcd3
- 10d ago Changed 9ca5df026acd
- 11d ago Changed · -13 lines · +9 tokens per session 65e14c8297b4
- 13d ago First seen · 54 lines · 70 tokens per session scan A 883c546dca4a
drive is a skill published in the GitHub repository juxt/claude-plugins (10 stars, last pushed yesterday), licensed MIT. It adds 166 tokens to every session and 1,691 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-04.
Other skills, from other repositories
github-copilot-upgrader
Use this to update the Github Copilot CLI/SDK.
workflow-patterns
Use this skill when implementing tasks according to Conductor's TDD workflow, handling phase checkpoints, managing git commits for tasks, or understanding the verification protocol.
meta-pre-commit-quality-gate
Run three quality gates (ruff + mypy + pytest) in parallel over the staged diff, then arbitrate a single BLOCK/APPROVE verdict. Use before committing changes locally when you want a comprehensive pre-commit gate beyond per-file linting — exactly the same gate set CI enforces.
verify-akashic-plugin
A plugin acceptance skill that commits a plugin, installs it into the running application, checks its status, and uses it once to confirm whether it works.
coverage-drop-investigator
Use when test coverage has dropped and you need to find which changes caused it and what tests to add. Traces coverage regressions to specific commits and files.
Changelog Test Mapper
Map changelog entries and release notes to affected test cases, ensuring every user-facing change has corresponding test coverage verification.